Operative Monitoring of the Extent of Dredging Plumes in Coastal

Document Sample
Operative Monitoring of the Extent of Dredging Plumes in Coastal Powered By Docstoc
					Journal of Coastal Research        SI 50          180 - 184            ICS2007 (Proceedings)          Australia        ISSN 0749.0208

Operative Monitoring of the Extent of Dredging Plumes in Coastal
Ecosystems Using MODIS Satellite Imagery
T. Kutser, L. Metsamaa, E. Vahtmäe and R. Aps
Estonian Marine Institute
University of Tartu, Tallinn
12618, Estonia


                       KUTSER, T., METSAMAA, L., VAHTMÄE, E. and APS, R., 2007. Operative monitoring of the extent of dredging
                       plumes in coastal ecosystems using MODIS satellite imagery. Journal of Coastal Research, SI 50 (Proceedings of
                       the 9th International Coastal Symposium), 180 – 184. Gold Coast, Australia, ISSN 0749.0208

                       MODIS band 1 imagery is the only satellite data with acceptable spatial resolution (250 m) and sufficient revisit
                       times (up to 4 times a day) potentially available for monitoring dredging plumes. The aim of this study was to
                       determine the potential use of MODIS band 1 imagery in an optically sophisticated environment such as the
                       Baltic Sea coastal waters for monitoring of total suspended matter in natural conditions and near dredging
                       activities. It was found that there is linear correlation between MODIS band 1 (620-670 nm) reflectance and the
                       total suspended matter concentration measured from water samples. The regression algorithm was used to
                       produce suspended matter concentration maps and monitor dredging activities. The results showed that in such
                       shallow water environments as the Estonian Coast resuspension caused by strong wind may increase the
                       suspended matter concentration to the levels similar to those occurring during dredging. Areas affected by wind
                       resuspension are much larger than the areas affected by dredging plumes. Therefore, the natural variability of
                       suspended matter concentration has to be taken into account when estimating environmental impact of dredging
                       activities on local ecosystems.

                       ADITIONAL INDEX WORDS: Remote sensing, MODIS, Suspended matter,Port dredging

                     INTRODUCTION                                       account expenses related to flying of an aircraft and the size of
   Dredging activities may cause major impact on coastal                impacted area that is usually relatively small.
ecosystems decreasing the amount of light available for                    There are two MODIS sensors on polar orbiting satellites Terra
photosynthesis and covering seagrass beds, coral reefs or fish          and Aqua. The swath of these sensors is such that in higher
spawning sites with a layer of sediments with potentially               latitudes each area gets imaged at least twice a day by both
disastrous consequences. On the other hand storms or tides also         satellites. It means that there are four images of a targeted site
cause resuspension of sediments. Therefore, one needs the               available in each cloud free day. This kind of revisiting time is
knowledge about variation of suspended sediment concentration in        usually more than adequate for monitoring dredging plumes.
natural conditions to assess the impact of dredging activities.         Spatial resolution of MODIS spectral bands is 1 km like the
Dredging may be needed in ports and waterways close to fish             spatial resolution of other ocean colour sensors (SeaWiFS,
spawning sites, coral reefs or other sensitive areas. In those cases    MERIS). This is not sufficient for coastal water monitoring.
there may be weather conditions when dredging has to be stopped         Especially taking into account that it is not recommended to use 2-
to avoid its most damaging effects. Operative monitoring methods        5 pixels nearest to the shore as those contain radiance reflected
are therefore needed to minimise harmful effects of dredging.           back from adjacent land not only from the water.
   One of the options would be using boats and water sampling for          MODIS has two bands with 250 m resolution. Those were
monitoring the extent of dredging plume and concentration of            designed for land, cloud and aerosol identifications and are in red
resuspended sediments within it. This, however, is expensive and        (620-670 nm) and near infrared (841-876 nm) spectral regions. It
time consuming. Remote sensing from aircraft (COLLINS and               is assumed that water leaving radiance is zero at those
PATTIARATCHI, 1984) or medium resolution (30 m) satellite               wavelengths due to strong absorption by water molecules and the
sensors (LINDELL et al., 1985) can provide detailed information         measured signal in these bands is only due to backscattering of
about the concentration of suspended matter in waterbodies.             light from atmospheric aerosols. However, it has been shown that
However, satellites with medium (30 m) to high (2.4 m) spatial          the water leaving radiance in MODIS bands 1 and 2 is not zero in
resolution do not have high enough revisit time to guarantee            coastal waters if there are cyanobacterial blooms (KUTSER, 2004;
operative monitoring of a targeted area. For example Landsat            KUTSER et al., 2006a,b; REINART and KUTSER, 2006), in optically
(30 m resolution) revisit time is 16 days. Airborne monitoring can      shallow waters (KUTSER et al., 2006c) and in waters with high
provide much more detailed and frequent overview of the extent          suspended sediment concentrations (DOXARAN et al., 2003).
of dredging plumes. However, using of airborne sensors may not          MILLER and MCKEE (2004) have shown that there is strong
be cost effective compared to in situ measurements taking into          correlation between the total suspended matter (TSM) and MODIS

                                           Journal of Coastal Research, Special Issue 50, 2007
                                     Operative monitoring of dredging plumes with MODIS satellite                                                                                                       181

                                                                       Satellite Data
                                                                       MODIS 250 m resolution data was downloaded from NASA web
                                                                       page http://disc.sci.gsfc.nasa.gov/data/datapool/MODIS/. MODIS
                                1                        2             band 1 reflectance images were geo-corrected, land and cloud
                                                                       areas were masked out. Atmospheric correction of theimages was
                                                                       performed using dark pixel method (CRANE, 1971) i.e. reflectance
                                                                       value of the darkest water pixel was subtracted from each image.

                                                                                                                                     RESULTS AND DISCUSSION
                                                                          In 2004 we used MODIS band 1 data to monitor dredging near
                                                                       Muuga port (#1 in Figure 1). Water samples were taken in 11
                                                                       locations nearly simultaneously with the satellite overpass.
                                                                       Correlation between the MODIS band 1 reflectance values and the
                                                                       total suspended matter concentrations in Muuga Bay is shown in
                                                                       Figure 2. The correlation is strong. However, the regression
                                                                       algorithm obtained (dashed trend line in Figure 2) shows that the
                                                                       minimum suspended matter concentrations can be 2.96 mg/l (if the
                                                                       reflectance value is zero). Suspended matter concentrations typical
Figure 1. Location of the ports under investigation. 1) Muuga port     for the open parts of the Baltic Sea are usually below 2 mg/l.
near Tallinn, 2) Sillamäe port near Estonian-Russian border.           Therefore, the suspended matter concentration was forced to be
                                                                       zero in case MODIS band 1 reflectance is zero (solid line in
                                                                       Figure 2) to get a more universal remote sensing algorithm for
                                                                       estimating TSM from MODIS band 1 data. The regression
band 1 imagery in waters with strong optical gradient (TSM from
                                                                       algorithm obtained by us differs from that obtained by MILLER and
almost 0 mg/l to 60 mg/l) like the Mississippi River plume in the
                                                                       MCKEE (2004) for Mississippi plume waters. Their algorithm was
clear waters of the Gulf of Mexico. The Baltic Sea waters are
relatively turbid and with high concentration of coloured dissolved
organic matter. However, the variability in TSM is much lower
than for example near the Mississippi plume. Therefore, using of
the MODIS band 1 data for mapping TSM content may be more                                                                 12
complicated in the Baltic Sea. MODIS band 1 imagery is the only
satellite data with acceptable spatial resolution and sufficient                                                          10
                                                                           T o ta l s u s p e n d e d m a tte r, m g /l

                                                                                                                                   y = 349.83x + 2.9663
revisit times potentially available for monitoring dredging plumes.                                                                     R2 = 0.8646
The aim was to study if it is possible to use this system in such an                                                      8
optically sophisticated environment as the Baltic Sea coastal
waters for monitoring of TSM in natural conditions and near                                                               6
dredging activities.

                           METHODS                                                                                        2
                                                                                                                                                                  y = 707.95x
                                                                                                                                                                  R2 = 0.5192

Study Site and In Situ Measurements                                                                                       0
   The study was carried out near two ports in the northern shore                                                              0    0.002     0.004       0.006    0.008        0.01   0.012   0.014   0.016
of Estonia: Muuga port near Tallinn and Sillamäe port near the                                                                                            MODIS band 1 reflectance
border with Russia (see Figure 1). Muuga is one of the largest
ports in the Baltic Sea but Sillamäe is a small recently built port.
Dredging in the ports is needed due to expanding of both of the          Figure 2. Correlation between MODIS band 1 reflectance and
ports, but periodical dredging is also needed due to continuous          total suspended matter in Muuga Bay in October 14, 2004.
sediment transport by currents.                                          Dashed line is the regression line and solid line is regression
   Surface water samples for total suspended matter measurements         line forced to go through point zero.
were collected in Muuga Bay in August 18, 2004 and in Narva
Bay close to Sillamäe port in September 13, 2006. Dried and pre-          A map of suspended matter concentration in Muuga and the
weighted 0.45 µm pore size filters were used to measure the total      surrounding bays is shown in Figure 3. The plume of the dredging
suspended matter concentrations from the water samples. 11             works near port of Muuga extends about 20 km north-west as seen
samples were collected in Muuga. The sampling stations were            in the map. The area most severely impacted by the dredging is
selected to cover the whole variability in suspended matter            not very large. Width of the plume is below 3 km in the widest
concentrations. The most turbid sample was taken a few tens of         place and is mainly less than 1 km for most of the plume. The
metres from the dredge; several samples were taken in the              highest concentrations of TSM remained below 9 mg/l even very
sediment plume and some samples outside the plume area where           close (few tens of metres) to the dredge according to our in situ
the suspended matter concentrations were equal to the natural          measurements. The suspended matter concentrations dropped to 4-
background values at the time of the measurements. Three               7 mg/l few hundred metres off the dredge according to our in situ
samples were collected near Sillamäe Port in September 13, 2006.       measurements and the concentration map obtained from MODIS

                                           Journal of Coastal Research, Special Issue 50, 2007
182                                                           Kutser et al.




                                                                                                               2 mg/l

Figure 3. Map of suspended matter concentration in Muuga Bay in October 14, 2004 obtained from MODIS band 1 data using the
regression algorithm developed using local in situ data (Equation 1).
                                                                      developed based on the Muuga Bay data slightly underestimated
image. Background TSM values for Most of the Muuga Bay in             TSM concentrations in the Narva Bay.
that particular day were around 2-3 mg/l according to the MODIS          Thus the remote sensing algorithm (Equation 1) enables
map.                                                                  estimates of suspended matter concentrations with certain
   The remote sensing algorithm obtained for Muuga Bay:               accuracy to be used in other sites beside the site it was developed
                                                                      for. However, collecting more TSM data simultaneously with
                  TSM=707.95*B1                            (1)        MODIS image acquisition is needed to determine whether or not
was used for monitoring dredging works near port of Sillamäe (#2      local TSM algorithms are needed or is it possible to modify the
in Figure 1) in September 2006. A time series of suspended matter     current algorithm to be suitable in different locations.
maps of the Narva Bay, where the port of Sillamäe is located, is         Results obtained in the Narva Bay showed that in natural
shown in Figure 4. Strong westerly wind (10-15 m/s) was               conditions the TSM concentrations may reach values similar to
observed during two days prior to the first clear day when the first  those in dredging plume. Moreover, the areas impacted by wind
MODIS image in our time series (Figure 4A) was acquired. It is        resuspension are significantly larger in our study area than those
seen that plume of resuspended sediments reaching west from the       impacted by human activity. Stormy days with wind speeds
Eastern shore of the Narva Bay has similar TSM concentrations         exceeding those recorded prior to our image time series occur
than the dredging plume area which is hardly noticeable in Figure     several (tens of) times a year. Flora and fauna have to be adapted
4A. The total area of each suspended sediment map in Figure 4 is      to such high variations in suspended matter concentrations if they
approximately 100x75 km. The most dense part of the                   still exist in this area. Assessing actual impacts of increased
resuspended sediment plume (shown in red and brown colours)           suspended matter concentration on local ecosystem was not the
caused by strong wind reaches about 22 km offshore while the          topic of this study. However, knowledge about the extent of
dredging plume is about 3 km in size.                                 dredging plumes and the increase in suspended matter
   The dredging plume area becomes clearly observable two days        concentrations on the dredging area compared to natural
later (Figure 4B) when part of the resuspended (by wind)              conditions are needed before one can estimate impact of dredging
sediments has settled from the surface water layer. It took several   activities on local ecosystems. This study enabled estimates of the
days (Figure 4D) before most of the resuspended (by wind)             natural variability for Narva Bay and the extent of the area
sediments settled and the TSM values dropped to the values where      affected by storms and dredging near the main port in the region
they are in calm weather conditions. Movements of the dredging        of Sillamäe.
plume, due to changing wind direction, were seen in the time             There are several phenomena that may theoretically confuse
series of TSM maps from Narva Bay. There is a small area with         detecting and mapping of dredging plumes. Using a single spectral
high TSM values slightly west from the dredging plume seen in         band of a remote sensing sensor does not allow determination of
most maps in Figure 4. This is the area where dredged sediments       precisely what causes change in the measured signal. There are no
were dumped to the sea.                                               regions in the visible part of the spectrum where the remote
   Three water samples were taken for TSM analysis in September       sensing signal is affected by only one source. For example water
13, 2006. Correlation between the TSM values measured from            leaving signal in MODIS band 1 may be caused by three main
water samples and estimated from MODIS band 1 using the               reasons: high suspended matter concentration, bottom visibility in
Equation 1 was good (R2=0.85). However, the algorithm                 shallow water areas, and phytoplankton blooms.

                                          Journal of Coastal Research, Special Issue 50, 2007
                                   Operative monitoring of dredging plumes with MODIS satellite                                183

 A                                                                  B

 C                                                                  D

 E                                                                  F
Figure 4. Suspended matter concentration maps of Narva Bay in September 2006 (A) September 11; (B) September 13; (C) September
16; (D) September 17; (E) September 21; (F) September 22. Location of Sillamäe Port is indicated with arrows. Land and clouds are
masked white. Some of marine areas are white due to the colour scale which enables presenting the TSM variability better in the
dredging area. Colour scale is slightly varying from image to image due to the same reason. In general brown colours mean TSM
concentrations above 8 mg/l, red and yellow colours 4-8 mg/l, green 2-4 mg/l and white below 2 mg/l.

                                        Journal of Coastal Research, Special Issue 50, 2007
184                                                            Kutser et al.

   In the Baltic Sea, the bottom is usually visible in waters less                      LITERATURE CITED
than 5-6 m deep. It means that only the first couple of pixels from    COLLINS M. AND PATTIARATCHI C., 1984. Identification of
the shore contain some signal from the water bottom even in the           suspended sediment in coastal waters using airborne thematic
areas where water depth is increasing slowly. Masking out these           mapper data. International Journal of Remote Sensing. 4:635-
areas in suspended matter maps is not difficult.                          657.
    It is not possible to separate elevated water leaving signal in    CRANE, R.B., 1971. Preprocessing techniques to reduce
MODIS band 1 caused by phytoplankton blooms from those                    atmospheric and sensor variability in multispectral scanner
caused by high TSM. It has been shown (KUTSER et al., 2006a;              data. Procedings of the Seventh International Symposium on
REINART and KUTSER, 2006) that MODIS band 1 data can be used              Remote Sensing of Environment. 1345-1355.
for mapping cyanobacterial blooms. Cyanobacterial blooms               DOXARAN, D.; FROIDEFOND, J.-M. and CASTAING, P., 2003.
usually occur in the open Baltic Sea whereas dredging is taking           Remote-sensing reflectance of turbid sediment-dominated
place close to shoreline. Spatial signatures of these two events are      waters. Reduction of sediment type variations and changing
also different. Thus, phytoplankton blooms should not be a big            illumination conditions effects by use of reflectance ratios.
obstacle in mapping of dredging plumes with MODIS band 1 data.            Applied Optics. 42:2623-2634.
However, some precaution must be taken while estimating natural        KUTSER, T., 2004. Quantitative detection of chlorophyll in
TSM background values for a certain region before dredging                cyanobacterial blooms by satellite remote sensing. Limnology
activities as some of the TSM is due to phytoplankton. Periods of         and Oceanography. 49:2179-2189.
phytoplankton blooms have to be omitted from the analysis when         KUTSER, T.; METSAMAA, L.; VAHTMÄE, E., and STRÖMBECK, N.,
natural background values of suspended matter are estimated.              2006a. On suitability of MODIS 250 m resolution band data
   Remote sensing sensors can collect information only regarding          for quantitative mapping cyanobacterial blooms. Proceedings
surface layer of waterbodies. Thickness of this layer may vary            of Estonian Academy of Sciences. Biology. Ecology. 55:318-
from a few centimetres to tens of metres depending on water               328.
clarity. Concentration of particulate and dissolved material in the    KUTSER, T.; METSAMAA, L.; VAHTMÄE, E., and STRÖMBECK, N.,
top mixed layer of waterbodies is constant. It means that the depth       2006b. Monitoring cyanobacterial blooms by satellite remote
where remote sensing estimates the concentrations of water where          sensing. Estuarine Coastal and Schelf Science. 67:303-312.
constituents are valid is equal to the thickness of mixed layer        KUTSER, T.; VAHTMÄE, E., and MARTIN, G., 2006c. Assessing
(about 10 m in the Baltic Sea during summer period). Our                  suitability of multispectral satellites for mapping benthic
turbidity measurements show that sediments are well mixed from            macroalgal cover in turbid coastal waters by means of model
surface to bottom near dredges. Relatively heavy weight of                simulations. Estuarine, Coastal and Shelf Science. 67: 521-
mineral particles causes fast sedimentation (a few days as seen in        529.
Figure 4) which results in homogenous vertical distribution of         LINDELL, L.T., STEINVALL O. and JONSSON, H., 1985. Mapping of
suspended matter in the water column. It means that the extent of         coastal water turbidity using Landsat imagery. International
the suspended sediment plume seen in the water surface is equal to        Journal of Remote Sensing. 6:629-642.
the plume extent down to the bottom. This, however, is dependent       MILLER, R.L. and MCKEE, B.A., 2004. Using MODIS Terra 250
on the sediment type present in the dredging area. For example,             m imagery to map concentrations of total suspended matter in
there are very fine clay sediments in certain parts of the Estonian         coastal waters. Remote Sensing of Environment. 93:259-266.
coast. These sediments may stay for weeks in the water column          REINART, A. and KUTSER, T., 2006. Comparison of different
after dredging. Another effect that was observed with such fine             satellite sensors in detecting cyanobacterial bloom events in
sediments is that in calm weather they settle on the thermocline            the Baltic Sea. Remote Sensing of Environment. 102:74-85.
not on the bottom. A situation was observed where the surface
layer (seen by satellites) and bottom the layer were clear, but
turbidity was noted in a thin layer (about 1 m in our case) above
the thermocline and was several times more turbid. In such
situations the suspended matter can travel further than remote
sensors indicate. Some preliminary knowledge about the local
sediment type is needed to estimate if the remote sensing maps of
TSM are valid from the surface to bottom or just for the surface
(mixed) layer.

  MODIS band 1 (250 m spatial resolution) data can be used for
quantitative mapping of suspended matter in coastal waters despite
the band not being designed for that purpose. The study has shown
that MODIS band 1 data is an effective tool to map the extent of
plumes arising from port dredging activities both spatially and
  The results showed that in natural conditions suspended matter
values may reach those observed in a dredging plume. Areas
impacted by wind resuspension are usually much larger than those
impacted by dredging. Therefore, natural variability of local TSM
values has to be known to adequately estimate the impact of
dredging activities on local ecosystem.

                                           Journal of Coastal Research, Special Issue 50, 2007

Shared By:
Description: Operative Monitoring of the Extent of Dredging Plumes in Coastal ...